Synapses are the sites of information processing in the mammalian brain. While generation and persistence of synapses during learning make them potential substrate for circuit modification and memory storage, the molecular mechanisms underlying synaptic structural changes and synapse diversity remain to be elucidated. Dendritic spines are the postsynaptic sites for the majority of excitatory synapses in the brain. Using an innovative combination of in vivo two-photon imaging and retrospective Array Tomography, the goals of this proposal are to determine the molecular composition and local connectivity of learning-related synapses, and to reveal principles of circuit remodeling during learning. We propose three specific aims. Aim 1 examines the protein expression of individual spines (postsynaptic structures of excitatory synapses) during the process of synaptogenesis. Such expression patterns will be compared with those of preexisting stable spines to determine the molecular signature of new spines formed during learning. Aim 2 identifies the presynaptic partners for learning-associated new spines, and dissects how local circuits reweigh or rewire during learning. Aim 3 investigates how spine dynamics of pyramidal neurons from different cortical layers respond to motor learning, and determines if new spines formed during learning receive unique presynaptic inputs. Successful completion of the research will not only provide critical insights into the biology of synapse formation and diversity, but also offer neuroscientists a novel tool box to highlight new connections formed in a brain's recent past. Discovering how synapses remodel during learning will build a foundation for future investigation of how synaptic structure/function is altered by pathologies associated with learning defects.